An Economical AI for Diabetes Management

Lead Research Organisation: University College London
Department Name: Institute of Health Informatics


Motivation: Parkinson's disease is a debilitating degenerative movement disorder marked by tremor, rigidity, and bradykinesia, causing great disability. It is the fastest growing neurological disease (Dorsey and Bloem, 2018) and thus, there is an urgent need for disease-modifying treatment. However, so far, treatment is limited to the symptomatic.

The microbiome is an emerging candidate for treatment development. It is highly accessible, and any potential treatment would be easily scalable. Moreover, it introduces the option of preventative treatments, using the unique pre-clinical window of up to 20 years (Hawkes, Del Tredici and Braak, 2010). However, there is a fundamental problem with developing therapeutics based upon the gut microbiome; the sheer scale and complexity of the microbiome which includes one hundred trillion individual microbes and over 1,000 species interacting in non-linear ways.

To tackle this, the Sharma Lab has already used topological data analysis (TDA), an unsupervised ML technique which analyses the shape of the data as represented by a high-dimensional point cloud, and analysed a dataset of 1,151 microbiome samples (the 'atlas1006' dataset, see: Lahti et al., 2014). The method can specifically account for complex, high-dimensional, and noisy data and is thus ideal to uncover the non-linear relationships of the human microbiome (Carlsson, 2009; Liao et al., 2019). This was combined with a population-based approach to infer indirect association with PD.

The goal of the MRes/PhD project is to extend this work into larger datasets such as the GMRepo of 27k samples and connect it with supervised ML/DL methods for additional classification. This will enable in silico modeling of the microbiome - meaning manipulation of the microbiome network to find specific treatment targets and model response to drugs.


10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/S021612/1 01/04/2019 30/09/2027
2424027 Studentship EP/S021612/1 28/09/2020 30/09/2024 Eva-Katherine Lymberopoulos